Re: [RFC v4 PATCH 47/49] multi-process: Enable support for multiple devices in remote

[Jag Raman]

On 11/11/2019 11:15 AM, Stefan Hajnoczi wrote:
> On Thu, Oct 24, 2019 at 05:09:28AM -0400, Jagannathan Raman wrote:
> > @@ -93,7 +94,8 @@ static void process_config_write(MPQemuMsg *msg)
> >       struct conf_data_msg *conf = (struct conf_data_msg *)msg->data2;
> >   
> >       qemu_mutex_lock_iothread();
> > -    pci_default_write_config(remote_pci_dev, conf->addr, conf->val, conf->l);
> > +    pci_default_write_config(remote_pci_devs[msg->id], conf->addr, conf->val,
> > +                             conf->l);
> >       qemu_mutex_unlock_iothread();
> >   }
> >   
> > @@ -106,7 +108,8 @@ static void process_config_read(MPQemuMsg *msg)
> >       wait = msg->fds[0];
> >   
> >       qemu_mutex_lock_iothread();
> > -    val = pci_default_read_config(remote_pci_dev, conf->addr, conf->l);
> > +    val = pci_default_read_config(remote_pci_devs[msg->id], conf->addr,
> > +                                  conf->l);
> >       qemu_mutex_unlock_iothread();
> >   
> >       notify_proxy(wait, val);
>
> msg->id was read from a socket and hasn't been validated before indexing
> into remote_pci_devs[].

We see the common thread, w.r.t your concerns about security. Thanks for
pointing them out.

We will fix this and other similar issues in the future.

Thank you very much!
--
Jag

>